YouTube Transcript — Captions & AI Speech to Text avatar

YouTube Transcript — Captions & AI Speech to Text

Pricing

from $5.00 / 1,000 caption transcripts

Go to Apify Store
YouTube Transcript — Captions & AI Speech to Text

YouTube Transcript — Captions & AI Speech to Text

Get a YouTube transcript from any public video: its own captions for $0.005, or AI speech-to-text when it has none. Get text with timestamps, SRT/VTT subtitles, word timing, speaker labels and RAG-ready JSONL. Caption transcripts for videos up to 8 hours. No login, no cookies, no API key.

Pricing

from $5.00 / 1,000 caption transcripts

Rating

0.0

(0)

Developer

Sergey Andronik

Sergey Andronik

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

19 days ago

Last modified

Features

Limited permissions

Share

Get a YouTube transcript from any public YouTube video, with timestamps, as JSON, SRT or VTT. YouTube Transcript reads the video's own captions when it has them, which is the cheapest and fastest route at $0.005 per video. When a video has no captions, or you need speaker labels, it transcribes the audio with AI speech-to-text instead. Paste one link or a list of up to 1,000 and get one result per video: the full text, timed segments, subtitles, optional word timing and optional RAG-ready passages for search, summaries or an AI pipeline. No YouTube login, cookies or speech-provider API key are needed.

What you get for each YouTube video

  • Transcript text in text, plus transcript_llm with one line per segment for LLM prompts.
  • Where it came from: transcript_source is captions or asr, and caption_kind is manual (uploaded by the creator) or auto (generated by YouTube).
  • Timestamps per segment (t0, t1 in seconds), or per word with timestamps: "word".
  • SRT and VTT subtitles in the result and as downloads in the run's key-value store (transcript-000.srt, transcript-000.vtt, numbered by input position).
  • Readable paragraphs in paragraphs (start_s, end_s, speaker, text) and as transcript-000.txt and transcript-000.md downloads, at no extra charge. See Readable output.
  • Speaker labels (S1, S2, …) with diarize: true, through AI speech-to-text.
  • RAG passages with source URL and original-video seconds, inline as rag_chunks and as a transcript-000.rag.jsonl download, with exportRag: true.
  • Part of a video: startTime and endTime in seconds; timestamps stay relative to the full video.
  • Billing per item: every row shows which events were charged.

How to get a YouTube transcript

  1. Paste your video URLs. Watch links, youtu.be short links, Shorts, live replays and embed links work. For a whole channel or playlist, use channelUrls instead.
  2. Pick the transcript source. Keep Auto (captions first, AI speech-to-text when there are none), or choose Captions only or AI speech-to-text only.
  3. Add options if you need them: a language, word timestamps, speaker labels, RAG passages or a time range.
  4. Set a run budget in the run options, then start the run. Each URL becomes one row in the Dataset. Export it as JSON, CSV or Excel, or read it through the API.

Input

{
"urls": ["https://www.youtube.com/watch?v=zjkBMFhNj_g"],
"transcriptSource": "auto",
"timestamps": "segment"
}
FieldWhat it does
urlsUp to 1,000 public YouTube video URLs. Required unless you set datasetId.
datasetIdRead URLs from another run's dataset. See Chain from another Actor.
datasetUrlField, maxItemsThe dataset field that holds the URL (detected when empty), and how many rows to read (default 100, at most 1,000).
transcriptSourceauto (default): captions first, AI speech-to-text when there are none. captions: captions only; a video without a usable caption track returns an error and is not charged. asr: always AI speech-to-text.
languageOptional language code such as en, es or pt-BR. With captions it picks the caption track; with AI speech-to-text it guides recognition.
timestampsnone (text only: no segments, words, SRT or VTT), segment (default) or word. Word timing comes from YouTube's auto-generated captions when they carry it, otherwise from AI speech-to-text.
diarizeSpeaker labels. Needs AI speech-to-text and is billed at the speaker-label rate.
exportRagAdds rag_chunks and a .rag.jsonl download. Needs segment or word timestamps.
startTime, endTimeTranscribe only this part of the video, in seconds.

Start from a channel or playlist

Put channel or playlist URLs in YouTube channels and playlists (channelUrls) to transcribe their newest videos:

{
"channelUrls": ["https://www.youtube.com/@TED", "https://www.youtube.com/playlist?list=PLe3qqz7JaEDM"],
"maxItemsPerSource": 10,
"transcriptSource": "captions"
}
  • Channel URLs: /@handle, /channel/UC…, /c/… and /user/…, optionally with a /videos, /shorts or /streams tab. Without one, Channel tab (channelTab, default videos) decides. Playlist URLs: /playlist?list=…, in playlist order.
  • Each row adds source_url, source_type (youtube_channel or youtube_playlist) and source_title.
  • publishedAfter (YYYY-MM-DD) is approximate on YouTube: channel and playlist pages show only a relative age such as "3 weeks ago", so videos up to one such unit older can be included. Shorts have no date at all, so publishedAfter cannot be combined with the Shorts tab.
  • publishedAfter filters only the entries that were listed. A channel tab lists newest first, maxItemsPerSource deep, so it finds every recent video. A playlist keeps its creator's order, which can be oldest first; with publishedAfter it is listed up to 200 entries deep, and the first maxItemsPerSource entries on or after the date are kept. Newer videos further down a longer playlist are not found.
  • The listing request goes from Apify's servers directly to YouTube, not through the residential proxy that video data uses.
  • Listed videos follow transcriptSource like any other: captions first by default.
  • maxItemsPerSource (default 20, at most 200) limits how many of the newest items each source contributes. A run holds at most 1,000 items in total, counting urls and dataset rows.
  • Each listed item is processed and charged exactly like a URL you submit. Listing itself is free.
  • The example link in the URL list is processed and charged too: remove it when you want only listed items.
  • Listed items run after urls and dataset rows, source by source in input order; an item already in the run is skipped.
  • A source that cannot be listed becomes one error row with source_url and an error naming it (for example channelUrls[1]: …), and it is not charged. The other sources continue. If no source lists anything, the run fails before any charge.
  • The listing is saved when the run starts, so a restarted run keeps the same items even if the source changed meanwhile.

Scheduled runs: only new items

Turn on Skip items processed by earlier runs (skipPreviouslyProcessed) and run the Actor on a schedule to get only what is new since the last run.

  • The Actor keeps a key-value store named polyglot-processed-youtube in your Apify account, or polyglot-processed-youtube--<memoryKey> when you set Memory name (memoryKey), so separate schedules can keep separate memories. Other Actors keep their own stores, even with the same memoryKey. It records each item after the item is delivered and charged; failed items are tried again next time.
  • It applies to every URL in the run: listed ones, dataset rows and the items listed from sources.
  • When nothing is new, the run ends without any charge. Only error rows appear then, for sources or dataset rows that could not be used.
  • If the store cannot be opened or read, the run stops before any charge, and the next scheduled run tries again. If an item's record cannot be written, the item is still delivered and the log says so; a later run may process and charge it again.
  • Delete the store in Apify Console (Storage → Key-value stores) to start over. Do not run two runs with the same memory at the same time: both would process the same new items.

Chain from another Actor

Run a YouTube scraper, such as YouTube Scraper by this publisher first, then paste the dataset ID of that run into Dataset ID (datasetId) to transcribe every one of its YouTube videos:

{ "datasetId": "<dataset ID from the other run>", "maxItems": 200 }
  • The Actor reads up to maxItems rows (default 100) with your own run's Apify token and takes each row's URL from webVideoUrl, url, postUrl, videoUrl, inputUrl or link, or from the field you name in datasetUrlField.
  • Dataset URLs run after any urls you list, in dataset order; a dataset URL that is already in the run is skipped. A run takes at most 1,000 URLs in total.
  • A row without a usable URL becomes an error item with its dataset_row index, and it is not charged. Items from the dataset also carry dataset_row, so you can join the results back to the source rows.
  • The example link in the URL list is processed and charged too: remove it when you want only the dataset's items.

Output example

An illustrative result, built with this Actor's caption parser from the real English caption track of "Me at the zoo" (19 seconds). It is not a recorded Apify run; the SRT and VTT fields are shortened.

{
"url": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
"transcript_source": "captions",
"caption_kind": "manual",
"text": "All right, so here we are, in front of the elephants the cool thing about these guys is that they have really... really really long trunks and that's cool (baaaaaaaaaaahhh!!) and that's pretty much all there is to say",
"language": "en",
"duration_s": 19,
"segments": [
{ "t0": 1.2, "t1": 3.36, "text": "All right, so here we are, in front of the elephants" },
{ "t0": 5.318, "t1": 7.974, "text": "the cool thing about these guys is that they have really..." },
{ "t0": 7.974, "t1": 12.616, "text": "really really long trunks" },
{ "t0": 12.616, "t1": 14.367, "text": "and that's cool" },
{ "t0": 14.421, "t1": 15.733, "text": "(baaaaaaaaaaahhh!!)" },
{ "t0": 16.881, "t1": 18.881, "text": "and that's pretty much all there is to say" }
],
"srt": "1\n00:00:01,200 --> 00:00:03,360\n…",
"billing": {
"status": "paid",
"events": [{ "event": "caption-transcript", "requested_count": 1, "charged_count": 1, "state": "confirmed" }]
}
}

Results also contain vtt, transcript_llm, source_duration_s, start_time_s and end_time_s. Results from AI speech-to-text also contain engine and usage. A URL that fails returns a row with url and error. Very large results may carry result_record_key instead of inline text; read that key from the run's key-value store.

Video data in transcript rows

Every caption transcript, and every AI speech-to-text row, also carries the video's data at no extra charge. Caption rows and rows that followed a caption attempt take it from that attempt's lookup. Other speech-to-text rows (transcriptSource: asr, speaker labels) take it from the audio download's own video info, with no extra request: video_id, title, author_name, author_id, author_url, author_username, published_at, upload_date, view_count, like_count, comment_count, thumbnail_url, tags, categories, availability and live_status. The names match YouTube Video Scraper. A field is left out when YouTube does not show it, and all of them are left out when the download's video info was reduced or unreadable (the transcript is still returned). The description is not included; use YouTube Video Scraper for it.

Two more fields come from the same lookup:

  • video_chapters: the chapters the creator defined, as [{ "title", "start_s", "end_s" }]. Absent when the video has none. These are not the AI chapters of the text add-ons.
  • available_caption_languages: the caption tracks YouTube lists in the video's own languages, as [{ "language", "kind": "manual" | "auto", "name" }]. YouTube's machine translations of those tracks are not listed, and the list stops at 100 entries. A row for a video without a caption track in the language you asked for (captions mode, not charged) still carries this list and video_chapters, so you can see which languages exist.

Readable output

Every transcript row has paragraphs: the transcript in order, grouped into blocks you can read. A paragraph starts at a speaker change, after a pause of 2 seconds or more, and before it would pass 800 characters or 90 seconds. Once it holds 400 characters, the next sentence end closes it. No text is dropped or reordered, and a segment is never split in timed output. Each paragraph is { "start_s", "end_s", "speaker", "text" }; speaker appears only with diarize: true, and with timestamps: "none" there are no times and paragraphs are cut at sentence ends by length alone. A caption transcript is grouped the same way as AI speech-to-text. Rows without speech have no paragraphs.

The same paragraphs are saved to the run's key-value store, numbered by input position:

  • transcript-000.txt: paragraphs separated by blank lines, with S1: style speaker labels.
  • transcript-000.md: a title (the video title), a list with the source URL, language and duration (and the channel name), then each paragraph with its [mm:ss] (or [h:mm:ss]) start time and bold speaker label.

When a row would pass the Dataset size limit, paragraphs is left out of the row first (the text stays) and the downloads still hold it. Both downloads are free, like SRT and VTT.

How YouTube data is fetched

  • On Apify, the video's metadata and its caption track are requested through Apify's residential proxy, with a traffic limit for each video. That cost is included in the per-video price.
  • The Actor uses no YouTube login and no cookies.
  • For AI speech-to-text, the audio is downloaded from Apify's servers without that proxy. YouTube refuses some of these downloads.
  • When YouTube blocks a request or the video is unavailable, the item returns an error. A failed caption request or a failed download is not charged. When YouTube returns incomplete video data, the item fails without a charge and without falling back to AI speech-to-text; retry it later.

Length limits

RequestCaption transcriptsLonger videos
Segment timestamps or noneUp to 8 hourscaptions: an error, not charged. auto: AI speech-to-text, billed at its prices.
Word timestampsUp to 2 hourscaptions: an error, not charged. auto: AI speech-to-text, billed at its prices.

AI speech-to-text handles recordings up to 4 hours. A longer video fails after its audio download and is not charged. Selecting a time range still downloads the whole audio.

How much does a YouTube transcript cost?

Pay per event, with no subscription. These are the Free-plan prices; paid Apify plans get lower per-minute prices for AI speech-to-text. The Pricing tab shows the current prices for your plan.

EventPriceWhen
Run setup$0.005Once per run.
Caption transcript$0.005Per video whose transcript comes from its captions, at any length up to the limits above.
Audio download for AI speech-to-text$0.02Per video whose audio was downloaded. Not charged for caption transcripts.
Audio minute$0.01Per started minute of AI speech-to-text, after it succeeds.
Diarized audio minute$0.02Per started minute with speaker labels, instead of the audio minute.

Worked examples:

Video1 URL in a run100 URLs in a run
With captions, any length$0.005 + $0.005 = $0.010$0.005 + 100 × $0.005 = $0.505
No captions, 1 minute$0.005 + $0.02 + $0.01 = $0.035$0.005 + 100 × $0.03 = $3.005
No captions, 3 minutes$0.005 + $0.02 + 3 × $0.01 = $0.055$0.005 + 100 × $0.05 = $5.005
No captions, 10 minutes$0.005 + $0.02 + 10 × $0.01 = $0.125$0.005 + 100 × $0.12 = $12.005

With speaker labels, a 10-minute video costs $0.005 + $0.02 + 10 × $0.02 = $0.225.

  • In auto mode you pay the caption price only for videos that have usable captions, and the AI speech-to-text prices for the rest.
  • Minutes are rounded up once per video, and silence counts.
  • A video with no detected speech (only music, silence or noise) returns no_speech: true with empty text and is not charged minutes, up to 30 such minutes per run; after that its minutes are charged. The audio download fee still applies.
  • A successful audio download stays charged if the transcription fails afterwards. A download that is not playable audio, or is longer than 4 hours, is not charged.
  • Apify's own small run-start event is separate and is shown on the Pricing tab.

Use it from the API

Start a run with the Apify API client for JavaScript and read the results:

import { ApifyClient } from 'apify-client'
const client = new ApifyClient({ token: process.env.APIFY_TOKEN })
const run = await client.actor('andronixmd/youtube-transcript').call(
{ urls: ['https://www.youtube.com/watch?v=zjkBMFhNj_g'], transcriptSource: 'captions' },
{ maxTotalChargeUsd: 0.1 },
)
const { items } = await client.dataset(run.defaultDatasetId).listItems()
for (const item of items) console.log(item.url, item.error ?? item.text)

Or start one run over HTTP:

curl -X POST \
-H "Authorization: Bearer $APIFY_TOKEN" \
-H 'Content-Type: application/json' \
-d '{"urls":["https://www.youtube.com/watch?v=zjkBMFhNj_g"]}' \
'https://api.apify.com/v2/actors/andronixmd~youtube-transcript/runs?maxTotalChargeUsd=0.1'

The Actor also works from n8n and other automation tools through Apify's integrations.

Use it from an AI agent (MCP)

Connect your agent to https://mcp.apify.com. It can find this Actor with search-actors, or call it by name, andronixmd/youtube-transcript, with call-actor. Use fetch-actor-details first for the input schema and pricing. Every call starts a new paid run, so always set callOptions.maxTotalChargeUsd and keep the urls list short:

{ "actor": "andronixmd/youtube-transcript", "input": { "urls": ["https://www.youtube.com/watch?v=zjkBMFhNj_g"], "transcriptSource": "auto", "timestamps": "segment" }, "callOptions": { "maxTotalChargeUsd": 0.1 } }

call-actor returns the run status and storage IDs, not the results. Read the rows with get-dataset-items and the run's dataset ID. The fields to use are title, text, paragraphs and error.

With an Apify account, use your API token or MCP OAuth. Without one, an agent can buy a prepaid token from Apify AGI over x402 or MPP and use it as the Bearer token. It starts at $1, the balance is a hard spending cap, and the token expires after 14 days. This is experimental; see agentic payments.

Limitations

  • Public videos only. Private, members-only, age-restricted, region-blocked and DRM-protected videos, and anything that needs a login, are not supported.
  • Videos, channels and playlists. Search URLs are not accepted. One URL in a wrong format rejects the whole input before the run starts. Channel listing from Apify's servers has not been tested at scale; a channel that cannot be listed becomes an error row that is not charged.
  • Not every video works. YouTube can block requests or serve incomplete data, so some URLs return an error row. Check each row's error and billing fields and retry failed URLs in a later run.
  • Restarted runs. If a run is restarted or resurrected, the remaining videos fail without a charge; start a new run for them. A run that Apify moves to another server continues normally. After 20 videos in a row fail without a charge, the run stops requesting YouTube data and the rest fail the same way.
  • Live streams have no complete caption track until they end.

FAQ

Captions or AI speech-to-text: which should I use? Captions are cheaper and quicker, and exact when the creator wrote them. YouTube's auto-generated captions can contain errors, and caption_kind tells you which you got. AI speech-to-text works from the audio, so it covers videos without captions and gives speaker labels. Auto uses captions where they exist and AI speech-to-text elsewhere. It also uses AI speech-to-text when you ask for speaker labels, or for word timestamps on a caption track without word timing.

Which languages are supported? For captions, any language the video has a caption track in. Without language, the Actor picks the track in the video's spoken language, else English. If YouTube does not say which language is spoken and the creator captions include neither English nor a single track, set language to choose one. The result's language names the track used. For AI speech-to-text, the language is detected automatically, and support depends on the speech engine.

Can I download YouTube subtitles as SRT? Yes. With segment or word timestamps (segment is the default), each result contains srt and vtt, and the files are saved in the run's key-value store.

Can I get transcripts of a whole channel or playlist? Yes. Put it in channelUrls and set maxItemsPerSource (up to 200 per source). With skipPreviouslyProcessed on a schedule, each run picks up only new uploads.

Do you store my transcripts or audio? Results stay in your Apify run storage, under your account's access and retention settings. With AI speech-to-text, the audio is sent to the speech-to-text provider. With captions, no audio is processed. Only submit videos you are allowed to process.